In this second article from our series of articles covering the overall RPA Implementation Lifecycle, we will zoom into the Plan phase and discuss in particular about the Process Identification activity in this phase.
Process Identification Approaches
Identifying the right process or task to automate as part of your RPA implementation initiatives is an activity of high importance, that has a significant bearing on the success of your RPA program. We will explore two distinct approaches in identifying processes for automation that are complementary to each other-
- Identifying existing processes with Process Discovery,
- Identifying new automation ideas with Crowdsourcing
Before we look at these approaches to process identification, it is important to first understand the two broader RPA implementation approaches to which these can be aligned.
RPA Implementation Approaches – Top Down and Bottom Up
In this era of disruption as organisations age and grow, they go through periods of both incremental and discontinuous change. Often stable growth periods are followed by periods of substantial market and technological instability, that necessitates companies to abandon existing business practices and take up significant change to survive through such periods.
Stable growth periods call for incremental innovation needing existing business processes be made faster, better and cheaper by improving and optimising them further. Exploiting RPA and associated existing proven automation technologies can act as a key strategic lever in achieving this.
Uncertain growth periods instead demand radical innovation measures to transform existing business processes or develop new ones. Exploring and adopting newer disruptive automation technologies such as Machine Learning, Computer Vision and Natural Language Processing is increasingly becoming a key to differentiate and survive through disruption.
Achieving these two levels of innovation with RPA and automation requires different approaches, competencies, operating models, incentives and culture. While incremental innovation with automation can be achieved by a top-down approach, radical innovation with automation demands a bottom-up approach.
The different aspects of these two fundamental approaches are presented in the table1 below.
Align your Process Identification initiatives with your RPA Implementation Approach
From a process identification standpoint, a top-down RPA approach for existing stable processes is better supported with Process Discovery initiatives. Process Discovery which is originally a subset of Process Mining is now also available in a more tailored form for RPA initiatives.
Process mining is all about extracting and analysing data from the events logs of the information systems we use to perform business as usual. The event log data is at the core of having to draw insights about existing processes and to model them with process mining. The big assumption here is that each step and each activity performed on a system has been logged.
Although very effective at times to identify and capture process models for automation, there can be a number of issues2 with this technique when applied to identify processes for automation such as –
- Noise– Additional unwanted steps captured in the logs
- Missed steps– that does not support logging
- Duplicate steps– that resemble the same process or its part
- Concurrent steps– that are occurring simultaneously
- Variations– when the same process steps are performed by different personnel
Some of these issues, such as noise or variations can now be solved by applying machine learning algorithms. However, given the severity of these issues in certain situations, for example, missed steps in case of legacy applications and the amount of effort needed to resolve them. A newer form of process discovery has emerged, which instead of relying on the system event logs, records the user actions on a computer screen in terms of keystrokes, and mouse movements, etc. The information generated from the user actions this way is then used to model the process. There may still be noise or variations in process discovery models generated by visual recording needing further work.
Nonetheless, process discovery either in its traditional sense (the use of event logs) or the use of visual recording is increasingly being applied in RPA implementations, firstly to identify and generate a steady pipeline of candidate processes to assess for automation suitability. Secondly to automatically generate RPA workflow models to use for automation. Most RPA product vendors are now providing process discovery capabilities out of the box or allowing technology integrations into specialized 3rd party tools.
A Bottom-up approach to RPA, to transform existing processes or define new ones, requires companies to undertake innovation in their automation initiatives by Capturing, Nurturing, Applying and Institutionalising new automation ideas. This can be achieved by applying the idea of a knowledge brokering cycle in the context of automation3.
- Crowdsource Automation Ideas– New innovative automation ideas, that apply newer AI and Digital technologies such as ML, Computer Vision, NLP, IoT, Mobile, AR, etc, are crowd-sourced from automation enthusiasts across the organisation.
- Store and share Ideas– these ideas are then published on a knowledge management platform in the form of an internal automation components library. This internal automation library is then made visible and accessible across the organisation for re-use. Ideas are showcased and spread across the organisation.
- Apply ideas to solve problems– Captured ideas are tried by individuals involved in different technology projects across the organisation to solve the existing problems.
- Generate value from new ideas– Automation Ideas that work are tested, improved and applied in real-life business situations through a highly agile, decentralised, iterative process.
1 Sharma,D. (2020) Top-down and/or Bottom-up Automation?. https://www.linkedin.com/pulse/top-down-andor-bottom-up-automation-deepak-sharma/
2 Thaduri et al. (2014) Process Mining for Maintenance Decision Support: Performance and Safety Management. doi: 10.1007/978-981-10-7323-6_23
3 Sharma,D. (2020) Scale your Automation Initiatives by building an Automation Ideas Factory. https://www.linkedin.com/pulse/scale-your-automation-initiatives-building-ideas-factory-sharma/